Attributes’ Importance for Zero-Shot Pose-Classification Based on Wearable Sensors
This paper presents a simple yet effective method for improving the performance of zero-shot learning (ZSL). ZSL classifies instances of unseen classes, from which no training data is available, by utilizing the attributes of the classes. Conventional ZSL methods have equally dealt with all the avai...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2018-08-01
|
| Series: | Sensors |
| Subjects: | |
| Online Access: | http://www.mdpi.com/1424-8220/18/8/2485 |
| _version_ | 1818036718658387968 |
|---|---|
| author | Hiroki Ohashi Mohammad Al-Naser Sheraz Ahmed Katsuyuki Nakamura Takuto Sato Andreas Dengel |
| author_facet | Hiroki Ohashi Mohammad Al-Naser Sheraz Ahmed Katsuyuki Nakamura Takuto Sato Andreas Dengel |
| author_sort | Hiroki Ohashi |
| collection | DOAJ |
| description | This paper presents a simple yet effective method for improving the performance of zero-shot learning (ZSL). ZSL classifies instances of unseen classes, from which no training data is available, by utilizing the attributes of the classes. Conventional ZSL methods have equally dealt with all the available attributes, but this sometimes causes misclassification. This is because an attribute that is effective for classifying instances of one class is not always effective for another class. In this case, a metric of classifying the latter class can be undesirably influenced by the irrelevant attribute. This paper solves this problem by taking the importance of each attribute for each class into account when calculating the metric. In addition to the proposal of this new method, this paper also contributes by providing a dataset for pose classification based on wearable sensors, named HDPoseDS. It contains 22 classes of poses performed by 10 subjects with 31 IMU sensors across full body. To the best of our knowledge, it is the richest wearable-sensor dataset especially in terms of sensor density, and thus it is suitable for studying zero-shot pose/action recognition. The presented method was evaluated on HDPoseDS and outperformed relative improvement of 5.9% in comparison to the best baseline method. |
| first_indexed | 2024-12-10T07:15:24Z |
| format | Article |
| id | doaj.art-332e1283f5c94624b413e54f3cf91e87 |
| institution | Directory Open Access Journal |
| issn | 1424-8220 |
| language | English |
| last_indexed | 2024-12-10T07:15:24Z |
| publishDate | 2018-08-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Sensors |
| spelling | doaj.art-332e1283f5c94624b413e54f3cf91e872022-12-22T01:57:57ZengMDPI AGSensors1424-82202018-08-01188248510.3390/s18082485s18082485Attributes’ Importance for Zero-Shot Pose-Classification Based on Wearable SensorsHiroki Ohashi0Mohammad Al-Naser1Sheraz Ahmed2Katsuyuki Nakamura3Takuto Sato4Andreas Dengel5Research & Development Group, Hitachi, Ltd., Tokyo 185-8601, JapanGerman Research Center for Artificial Intelligence (DFKI), 67663 Kaiserslautern, GermnayGerman Research Center for Artificial Intelligence (DFKI), 67663 Kaiserslautern, GermnayResearch & Development Group, Hitachi, Ltd., Tokyo 185-8601, JapanResearch & Development Group, Hitachi, Ltd., Tokyo 185-8601, JapanGerman Research Center for Artificial Intelligence (DFKI), 67663 Kaiserslautern, GermnayThis paper presents a simple yet effective method for improving the performance of zero-shot learning (ZSL). ZSL classifies instances of unseen classes, from which no training data is available, by utilizing the attributes of the classes. Conventional ZSL methods have equally dealt with all the available attributes, but this sometimes causes misclassification. This is because an attribute that is effective for classifying instances of one class is not always effective for another class. In this case, a metric of classifying the latter class can be undesirably influenced by the irrelevant attribute. This paper solves this problem by taking the importance of each attribute for each class into account when calculating the metric. In addition to the proposal of this new method, this paper also contributes by providing a dataset for pose classification based on wearable sensors, named HDPoseDS. It contains 22 classes of poses performed by 10 subjects with 31 IMU sensors across full body. To the best of our knowledge, it is the richest wearable-sensor dataset especially in terms of sensor density, and thus it is suitable for studying zero-shot pose/action recognition. The presented method was evaluated on HDPoseDS and outperformed relative improvement of 5.9% in comparison to the best baseline method.http://www.mdpi.com/1424-8220/18/8/2485zero-shot learningwearable sensorIMUpose classificationaction recognitiontime-seriesCNN |
| spellingShingle | Hiroki Ohashi Mohammad Al-Naser Sheraz Ahmed Katsuyuki Nakamura Takuto Sato Andreas Dengel Attributes’ Importance for Zero-Shot Pose-Classification Based on Wearable Sensors Sensors zero-shot learning wearable sensor IMU pose classification action recognition time-series CNN |
| title | Attributes’ Importance for Zero-Shot Pose-Classification Based on Wearable Sensors |
| title_full | Attributes’ Importance for Zero-Shot Pose-Classification Based on Wearable Sensors |
| title_fullStr | Attributes’ Importance for Zero-Shot Pose-Classification Based on Wearable Sensors |
| title_full_unstemmed | Attributes’ Importance for Zero-Shot Pose-Classification Based on Wearable Sensors |
| title_short | Attributes’ Importance for Zero-Shot Pose-Classification Based on Wearable Sensors |
| title_sort | attributes importance for zero shot pose classification based on wearable sensors |
| topic | zero-shot learning wearable sensor IMU pose classification action recognition time-series CNN |
| url | http://www.mdpi.com/1424-8220/18/8/2485 |
| work_keys_str_mv | AT hirokiohashi attributesimportanceforzeroshotposeclassificationbasedonwearablesensors AT mohammadalnaser attributesimportanceforzeroshotposeclassificationbasedonwearablesensors AT sherazahmed attributesimportanceforzeroshotposeclassificationbasedonwearablesensors AT katsuyukinakamura attributesimportanceforzeroshotposeclassificationbasedonwearablesensors AT takutosato attributesimportanceforzeroshotposeclassificationbasedonwearablesensors AT andreasdengel attributesimportanceforzeroshotposeclassificationbasedonwearablesensors |